voice_audio_multi/model_specs/moss_voicegen.json

{
 "family": "moss_voicegen",
 "display_name": "MOSS-VoiceGenerator",
 "description": "MOSS voice design model: creates a speaker from a written instruction instead of a reference recording, then speaks the supplied text in that voice.",
 "category": "community",
 "status": "community",
 "tasks": [
  "design"
 ],
 "modes": [
  "offline"
 ],
 "languages": [
  "en",
  "zh"
 ],
 "capabilities": {
  "design": [
   "voice_design"
  ]
 },
 "options": {
  "request": [
   {
    "name": "min_frames",
    "type": "int",
    "required": false,
    "default": 0,
    "description": "Minimum generated audio frames per text chunk; nonpositive values derive a text-length-based floor."
   },
   {
    "name": "max_frames",
    "type": "int",
    "required": false,
    "default": 0,
    "description": "Generated audio-frame ceiling per text chunk; nonpositive values derive a text-length-based ceiling. Runtime raises it to at least the effective min_frames plus the number of codebooks."
   },
   {
    "name": "instruct",
    "type": "string",
    "required": false,
    "description": "Voice description: timbre, age, pace, emotion."
   },
   {
    "name": "text_temperature",
    "type": "float",
    "required": false,
    "default": 1.5,
    "description": "Text-token sampling temperature. Must be finite and positive when sampling."
   },
   {
    "name": "language",
    "type": "string",
    "description": "Optional full language name, such as English or Chinese; not an ISO language code.",
    "required": false
   },
   {
    "name": "seed",
    "type": "int",
    "description": "Generation seed; later text chunks add their chunk index. Parsed as signed 32-bit integer then converted to uint32; negative values wrap.",
    "required": false,
    "default": 0
   },
   {
    "name": "temperature",
    "type": "float",
    "description": "Audio-token sampling temperature. Must be finite and positive when sampling.",
    "required": false,
    "default": 1.5
   },
   {
    "name": "top_p",
    "type": "float",
    "description": "Audio-token nucleus sampling cutoff.",
    "required": false,
    "default": 0.6
   },
   {
    "name": "top_k",
    "type": "int",
    "description": "Audio-token top-k sampling limit.",
    "required": false,
    "default": 50
   },
   {
    "name": "repetition_penalty",
    "type": "float",
    "description": "Audio-token repetition penalty. Must be finite and positive when history is nonempty.",
    "required": false,
    "default": 1.1
   },
   {
    "name": "text_chunk_size",
    "type": "int",
    "description": "Maximum characters per generated text chunk; must be positive.",
    "required": false,
    "min": 1,
    "default": 200
   },
   {
    "name": "text_chunk_mode",
    "type": "enum",
    "description": "Text chunking policy.",
    "required": false,
    "preset": "text_chunk_mode_full",
    "default": "word_budget"
   }
  ],
  "session": [
   {
    "name": "weight_type",
    "type": "enum",
    "description": "Backbone weight storage. f16 is not supported.",
    "required": false,
    "values": [
     "native",
     "f32",
     "bf16",
     "q8_0"
    ],
    "default": "bf16"
   }
  ],
  "load": []
 },
 "runtime": {
  "tags": [
   "gguf"
  ]
 },
 "ui": {
  "recommended_package": "moss_voicegen_bf16_codec_f16_decode",
  "tags": [
   "Voice Design",
   "GGUF"
  ],
  "docs": [
   "docs/models/moss_voicegen.md",
   "docs/tts.md",
   "docs/gguf.md"
  ]
 },
 "package_defaults": {
  "download": {
   "kind": "huggingface_snapshot",
   "repo": "audio-cpp/audio.cpp-gguf",
   "revision": "main",
   "gated": false
  }
 },
 "packages": [
  {
   "id": "moss_voicegen_bf16_codec_f16_decode",
   "display_name": "MOSS-VoiceGenerator BF16 GGUF",
   "default": true,
   "format": "gguf",
   "precision": "bf16",
   "target_directory": "MOSS-VoiceGenerator-GGUF",
   "files": [
    "MOSS-VoiceGenerator-GGUF/moss_voicegen_bf16_codec_f16_decode.gguf"
   ],
   "strip_prefix": "MOSS-VoiceGenerator-GGUF"
  }
 ],
 "sources": [
  {
   "format": "gguf",
   "roots": {
    "model": ".",
    "weights": "$gguf"
   },
   "files": {
    "config": "model:config.json",
    "tokenizer_json": "model:tokenizer.json",
    "tokenizer_config": "model:tokenizer_config.json",
    "tokenizer_merges": "model:merges.txt",
    "audio_tokenizer_config": "model:audio_tokenizer/config.json"
   },
   "tensors": {
    "model_weights": {
     "source": "weights:",
     "prefix": "model_weights"
    },
    "audio_tokenizer_weights": {
     "source": "weights:",
     "prefix": "audio_tokenizer_weights"
    }
   }
  },
  {
   "format": "safetensors",
   "roots": {
    "model": ".",
    "audio_tokenizer": "audio_tokenizer"
   },
   "files": {
    "config": "model:config.json",
    "tokenizer_json": "model:tokenizer.json",
    "tokenizer_config": "model:tokenizer_config.json",
    "tokenizer_merges": "model:merges.txt",
    "audio_tokenizer_config": "audio_tokenizer:config.json"
   },
   "tensors": {
    "model_weights": "model:model.safetensors",
    "audio_tokenizer_weights": "audio_tokenizer:model.safetensors.index.json"
   }
  }
 ]
}